TERMS
Ten terms that carry my articles. Each definition stands on its own, dated, and points to the article in which it is derived.
10 terms · As of 2 September 2026
Scaling in the AI era means growing a B2B software company while AI changes cost structure, pricing and defensibility at the same time. Every answer costs compute, features become copyable, and value moves into data, workflow and distribution. Companies that scale put the economics in order first and build revenue on top.
Derived in: Scaling B2B SaaS in the AI era · As of 2 September 2026
By Defensible Scale I mean growth a competitor with model access can't simply rebuild. The method checks five layers, position, demand, monetisation, economics and compound, and shows which layer currently limits defensible growth. Investors read from it whether growth has substance or just speed.
Derived in: The Defensible Scale method on convios.com · As of 2 September 2026
An AI Operating Partner is an operating role alongside an investor or board: they're accountable for AI having a measurable effect on revenue, margin or cost in a portfolio company, with guardrails, a budget and one metric per initiative. The role differs from an AI consultant through result accountability inside the business.
Derived in: For PE investors and boards · As of 2 September 2026
The AI price ladder is a model that orders AI pricing in five steps: price per token, tariff mechanics with cache, batch and context tiers, price per capability, cost per solved task and price per outcome. With every step the cost risk moves further from the customer to the vendor. Buyers therefore negotiate the counting unit first and the price second.
Derived in: AI pricing models 2026: the AI price ladder · As of 2 September 2026
Cost per solved task ends up in the margin: machine cost of one attempt times the number of attempts, plus the error cost of the cases that weren't solved and the review cost of the people who sign off the result. The list price per token is only the first item on that bill, and often the smallest.
Derived in: A forty-ton truck for a loaf of bread: cutting AI costs · AI cost calculator · As of 2 September 2026
Software without a user interface is software that AI agents operate through APIs and standards such as the Model Context Protocol, with no human opening the screen. The value of the software then sits in the layer underneath: access logic, audit trail and business rules.
Derived in: Headless software · As of 2 September 2026
A permission layer defines, machine-readably, who may do what, and proves who did what. It carries four items: access logic over the data, a complete audit trail, codified business rules and the operating data from agent use. When agents skip the click path, the value of the software moves there.
Derived in: Headless software · As of 2 September 2026
The Rule of 40 adds growth rate and profit margin and expects at least 40. In the AI era compute cost runs inside delivery, and boards that enforce the rule under the old definition starve reinvestment in AI. The rule still holds when the AI cost per task is visible in the margin.
Derived in: Scaling B2B SaaS in the AI era · As of 2 September 2026
An AI operating system wraps AI around the whole company: the AI gets everything a new business partner would need to know, business model, team, strategy, tone of voice and processes, and from then on works like an informed colleague. The technical basis is agents that read, create and edit files.
Derived in: AI operating system for mid-sized companies on convios.com · As of 2 September 2026
Whether taking over a company pays off for a successor is what I call succession readiness (Übernahmewürdigkeit). It's decided years before the handover by recurring revenue, owner independence and documented knowledge. Each year around 114,000 German mid-sized owners plan to close, 109,000 look for a successor; the bottleneck now sits in the earnings.
Derived in: Succession and succession readiness on convios.com · As of 2 September 2026